OpenAI quietly presented its next generation of AI models, tentatively called « Astra », to stakeholders in Washington D.C. on August 1. This model family is designed for long-running, multi-agent tasks, a significant leap compared to the GPT-5.6 series launched in June and July 2026. Astra aims to coordinate multiple AI agents working together on complex, extended tasks. The preview was private and selective, with no benchmarks, performance metrics, or public release timeline disclosed. This multi-agent coordination capability could reshape how AI intersects with decentralized systems, with potential implications for the crypto sector.
Source: Read the original article
